This resolves:

     https://bugzilla.redhat.com/show_bug.cgi?id=851411
     https://bugzilla.redhat.com/show_bug.cgi?id=955500

The first problem was that virFileOpenAs was returning fd (-1) in one
of the error cases rather than ret (-errno), so the caller thought
that the error was EPERM rather than ENOENT.

The second problem was that some log messages in the general purpose
qemuOpenFile() function would always say "Failed to create" even if
the caller hadn't included O_CREAT (i.e. they were trying to open an
existing file).

This fixes virFileOpenAs to jump down to the error return (which
returns ret instead of fd) in the previously mentioned incorrect
failure case of virFileOpenAs(), removes all error logging from
virFileOpenAs() (since the callers report it), and modifies
qemuOpenFile to appropriately use "open" or "create" in its log
messages.

NB: I seriously considered removing logging from all callers of
virFileOpenAs(), but there is at least one case where the caller
doesn't want virFileOpenAs() to log any errors, because it's just
going to try again (qemuOpenFile()). We can't simply make a silent
variation of virFileOpenAs() though, because qemuOpenFile() can't make
the decision about whether or not it wants to retry until after
virFileOpenAs() has already returned an error code.

Likewise, I also considered changing virFileOpenAs() to return -1 with
errno set on return, and may still do that, but only as a separate
patch, as it obscures the intent of this patch too much.

The function takes pool name as argument. However,
it is not acceptable for it to be NULL. Hence, we
should check it and report error in case it is.

Added error checking to virEventRegisterDefaultImpl call for consistency.

In the domain-events example C code virEventRegisterDefaultImpl was being
called before virConnectOpen without first calling virInitialize. While this
code worked, it is incorrect. Adding a call to g_string_new prior to the call
to virEventRegisterDefaultImpl would cause the code to break. This fix will
help avoid unintentional misue of the API.

Relates to: Ret Hat Bugzilla - Bug 961155

s/vpu/vCPU/

'make check' fails since commit 470d5c46 on any system with dash
as /bin/sh, because '<<<' is a bash extension.  For example:

nwfilterschematest: 23: /home/eblake/libvirt/tests/schematestutils.sh: Syntax error: redirection unexpected

Also, there is no need to spawn a grep process when shell globbing
can do the same.

* tests/schematestutils.sh: Replace bashism and subprocess with a
faster and portable construct.
Signed-off-by: NEric Blake <eblake@redhat.com>

Some of the apparmor code files did not follow the normal
goto label naming pratices
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Update the DRIVER_SOURCE_FILES variable to reference the
other various XXX_SOURCES variables, instead of duplicating
the filename lists. This results in a bunch of extra files
being processed, but the test scripts can easily skip those
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

The previous update of method naming missed the ESX storage
backend files. Update them is that the driver impl methods
follow the naming of the public API but with s/vir/esx/
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

The virNWFilterTechDriver struct is an internal only driver
API with no public API equivalent. It should be skipped by
the 'check-driverimpls' test case
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Some methods in the udev interface driver used 'cleanup' as the
label for separate error codepaths. Change these to use 'error'
as required by coding standards
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

The udev interface driver did not follow standard naming
convention for goto labels.
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Older versions of libxml2 could not correctly parse certain
URIs. This causes test failures. There's nothing libvirt can
do about this, so disable the problem tests on old libxml2
versions
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

The PyDict_Next method on Python <= 2.4 used 'int' instead
of "Py_ssize_t" for the 'pos' parameter
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Commit 979e9c56 missed one case of providing the timestamp
parameter to virNetServerClientGetUNIXIdentity() when WITH_POLKIT0
is defined.

Print an error instead of crashing when a TPM device without
a backend is specified.

Add a test for tpm device with no backend, which should fail
with a parse error.

https://bugzilla.redhat.com/show_bug.cgi?id=961252

Currently, using an invalid XML in tests fails, because
the schema test expects all of them to be valid.

Treat files with -invalid.xml suffix as invalid and expect
them to fail validation.

The controller element supports non-disk controller types too.

https://bugzilla.redhat.com/show_bug.cgi?id=960958

Make the Xen domain stats / peek and node memory driver
methods unconditionally call the sub-drivers which are
guaranteed to be open.
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Unconditionally call into the XenD or XM drivers for autostart
handling, since they are guaranteed to be open

Make the Xen domain scheduler parameter methods directly
call into XenD or Xen hypervisor drivers
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Make the domain attach/dettach driver methods directly call
into either the XenD or XM drivers
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Make the domain define/undefine driver methods directly call
into either the XenD or XM drivers
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Directly call either the XenD or XM driver when starting
a persistent domain
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Directly call either XenD or the XM driver for handling
domain define operations.
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

All the migration code is done by the XenD subdriver which
can be assumed to always be present
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

The xenUnifiedDomainGetXMLDesc driver can assume that
the XM and XenD drivers are always present
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Unconditionally call the XenD APIs for save/restore, since that
driver will always be open.
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Make the xenUnifiedDomainGetInfo and xenUnifiedDomainGetState drivers
call the correct sub-driver APIs directly.
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Simplify the Xen memory limit driver methods to directly call
the most appropriate sub-driver
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

The xenGetHostname entry point in the xenUnifiedDriver table
was unused.
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Make xenUnifiedDomainGetOSType directly call either the
xenHypervisorDomainGetOSType or xenDaemonDomainGetOSType
method depending on whether the domain is active or not.
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>

Unconditionally call the xenDaemonDomainDestroyFlags API
since the XenD driver is always available.
Signed-off-by: NDaniel P. Berrange <berrange@redhat.com>